Adaptable Hypermedia with Web Standards and Tools Page 1 of 8 active web home || papers || posters Adaptable Hypermedia with Web Standards and Tools Lloyd Rutledge, Lynda Hardman, Jacco van Ossenbruggen and Dick C.A. Bulterman CWI P.O. Box 94079 1090 GB Amsterdam, The Netherlands E-mail: {lloyd,lynda,jrvosse,dcab}@cwi.nl The World Wide Web offers the promise, and more and more the reality, of a widely distributed and widely accessible digital library of related hypermedia data. In order for this data to be stored, accessed and played it should not be encoded as final form presentations, since these consume storage space and cannot easily be adapted to variations in presentation-time circumstances such as user characteristics, changes in end-user technology and the interactions of the user at presentation time. Instead, a more presentation independent approach needs to be taken that allows the dynamic generation of presentations that are derived from a presentation-independent description and are adapted to run-time circumstances. A collection of standards exists that address this possibility and its related challenges. The ISO standard HyTime (Hypermedia/Time-based Structuring Language) [10] specifies the representation of hypermedia documents in a presentation independent format. HyTime is used to a small but steady degree by private industry for large-scale text document collections. The widespread adoption of HyTime is inhibited because there are few widely available tools for processing it, there are few examples of its use, and few widely available systems exist that can generate hypermedia presentations from it. The ISO standard DSSSL (Document Style Semantics and Specification Language) [11] defines the transformation of electronic documents into formats that present them. SMIL (Synchronized Multimedia Markup Language, pronounced "smile") [9] is a new W3C (World Wide Web Consortium) recommendation for immediately presentable hypermedia documents distributed on the World Wide Web. DSSSL transforms documents encoded with Standard Generalized Markup Language (SGML), which is used as the foundation for defining both HyTime and SMIL. Because of this, DSSSL can encode the transformation of documents from HyTime to SMIL, and thus can encode the final presentation of documents stored in HyTime. The use of DSSSL with HyTime was recently made easier with the release of the second edition of HyTime, which contains new facilities for use with DSSSL. Publicly available tools exist that make the cooperative use of HyTime, DSSSL and SMIL for hypermedia digital libraries widely implementable. These standards and tools can be used together to create an environment that processes documents from stored hypermedia data into final presentations. In this presentation we present the Berlage environment design [15], [16], which applies these public standards and tools to create a storage-to-presentation hypermedia system.The design of Berlage is based on the Standard Reference Model for Intelligent Multimedia Presentation Systems (SRM-IMMPSs) [2], [15], which defines how the specification of automatically generated dynamic tailored presentations can be divided up into distinct modules. This dicussion of Berlage and the SRM-IMMPSs is illustrated with an example application about the city of Amsterdam, The Netherlands, called Fiets (Foundation for Interactive Electronic Touring Systems, or fiets {pronounced "feets"} [14], the Dutch word for "bicycle" and generally the preferred means of personal transportation in Amsterdam). Fiets provides a hypermedia interface to a digital library of media data regarding the city of Amsterdam, The Netherlands. The Fiets Hypermedia Application Hypermedia has typically been modeled in terms of its presentation. The resulting structure usually http://www.visualize.uk.com/conf/activeweb/proceed/pap18/ 12/6/2006 Adaptable Hypermedia with Web Standards and Tools Page 2 of 8 represents the timing of the media object presentations, the spatial layout of the visual media object displays, and the point-and-click navigational interface for the user. However, there is often a discernible structure to the concepts being conveyed by the presentation, and there can often be more than one means of using hypermedia presentation structure to convey these concepts and their underlying structure. Because of this, it is often beneficial for an author to model a document in terms of such an underlying conceptual structure and handle the mapping of this structure to the final presentation as a separate task. The Fiets application is used to explore the differences between time, space and hyperlinks in storage and presentation. The temporal structure of a document may have a direct mapping to the temporal structure of its presentation, but this is not always so. It could instead map to spatial or navigational presentation structure. Also, its mapping to the presentation structure may be altogether indirect or nonexistent. It is hoped that by exploring the mappings among these corresponding structures that the distinction between storage and presentation will be better understood and the potentials of maintaining such a distinction better utilized. The Fiets application is used as an example to illustrate these structural distinctions. Fiets consists of media objects about Amsterdam and metadata about these objects. These media objects consist of photographs of historic buildings along the Herengracht, one of Amsterdam's main canals. Fiets conveys the underlying conceptual structure of the Herengracht. The Herengracht has geographic structure: its buildings have locations, represented as street addresses. With its history, the Herengracht has a temporal structure: its buildings were constructed on particular dates. Further, the Herengracht buildings have structure consisting of semantic relationships. In Fiets, this is exemplified with detail images of building exteriors, which show particular portions of each building. Fiets illustrated the distinction between the time, space and hyperlinks of storage and of presentation by defining 9 different mappings of the possible combinations of each. These mappings are illustrated in Figure 1. Figure 2 shows a screen display from the generation from one such mapping: the "panorama" mapping of the spatial structure of the storage media and metadata (the street) to the spatial layout of the presentation (the screen display). http://www.visualize.uk.com/conf/activeweb/proceed/pap18/ 12/6/2006 Adaptable Hypermedia with Web Standards and Tools Page 3 of 8 Figure 1 Figure 2 The SRM-IMMPSs The SRM-IMMPSs is illustrated in Figure 3 and described below. http://www.visualize.uk.com/conf/activeweb/proceed/pap18/ 12/6/2006 Adaptable Hypermedia with Web Standards and Tools Page 4 of 8 Figure 3 The goal formulation component of the SRM-IMMPSs handles the initial interaction with the user that starts the presentation and establishes its goals. These goals are then passed to the control layer, which processes the goals to make sure they are met. While the breaking up of each goal into subgoals is determined by the content layer, the control layer is in charge of processing these subgoals to determine which is to be met next. This chosen subgoal is then communicated to the content layer, which generates the portion of the presentation that achieves it. In a paper-based society, a person reads a document to understand a particular body of knowledge. He or she can expect to be able to communicate about this topic with other people who have read the same paper document. The same common understanding should be expected for two people who have been presented with the same hypermedia document, no matter how much the document may have been adapted differently for each user. The content layer ensures that the recipient of the presentation for a given document has been properly shown all information within that document, regardless of how the other layers adapt the presentation. The design layer of the SRM-IMMPSs determines what means the final presentation uses to meet the goals established in the content layer. This layer makes decisions on the "look and feel" of the presentation, including what media objects should be selected and how they should be presented. The design layer also determines the spatial–temporal and navigational layout of the final presentation. The different style combinations of storage and presentation structure demonstrated by Fiets are shown in Figure 1. The realization layer translates the desires expressed by the design layer into a directly playable hypermedia format. This layer determines exactly what spatial coordinates and what timing can match the constraints given by the design layer. If no such detailed specifications are possible, it then communicates with the design layer to determine an acceptable alternative. Similar communication occurs when the initial desires of the design layer cannot be expressed in the directly playable hypermedia format. The two layers then decide what acceptable alternative can be expressed in the output format. In Fiets, the realization layer corresponds http://www.visualize.uk.com/conf/activeweb/proceed/pap18/ 12/6/2006 Adaptable Hypermedia with Web Standards and Tools Page 5 of 8 with the generation of the SMIL code that makes up the final presentation to the user. The presentation display layer is embodied by the particular player software that is called at presentation time to process a segment of code generated by the realization layer for presentation